YOUR AI AGENT CAN NOW UPGRADE ITSELF. Reef just launched /reefine: you describe what you want and it builds the change, tests it and ships a new version on its own. → Adds new modes → Tells you out loud → Integrates new tools → Works with Codex, OpenCode... Basically: your agent evolves just by talking to it. It's open source. Link below :)

the interesting number here isn't the 18%, it's the 700x. an eval you can afford to run on every response is a different product from one you run on a sample. that's the part that changes how you build.

Neobanks are great businesses that are easy to run. Today, you can create your own in 15 minutes on Whop: whop.com/blueprints/neobank

LA EDICIÓN DE VIDEO ACABA DE MORIR. Convirtieron a Claude en editor profesional… GRATIS. Arrastras tus archivos y ella sola corta, subtitula, corrige color y renderiza. Cero timeline. Cero clics. No asiste, reemplaza. Se llama Video Use, open source. Link en comentarios👇🏻

estudiantes de marketing viendo esto

ODYSSEY JUST LAUNCHED AGORA-2 AND IT'S A GAME ENGINE LEARNED BY AI → 20 humans and agents in one live world → 5x more players than Agora-1 → Multiple environments, not just one → Every action changes the game for all The bigger picture: this is how robots, self-driving cars and cyber agents will learn to cooperate and compete, safely in simulation. You can try it today

1B ARR in less than 2 years, 30M users and that’s only less than 0,5% of world population… imagine when AI goes mass adopted, Higgsfield is competing with industry giants and with way LESS time here CRAZY

I TRIED AMBIENT BY @vellum_ai TO BUILD A CAMPAIGN REPORT IN GOOGLE SHEETS Double-tapped Fn, shared my screen and asked "help me get the average CPM per account from this sheet" What happened: → Spotted views stored as text → Showed me on screen where to fix it → Dictated the formula, I pasted it with hold Fn About 4 minutes to a finished table. By hand it usually takes me 15 Not perfect: once it annotated the wrong column and I had to correct it by voice. It got it right away For screen work where you'd normally take screenshots and paste them into a chat, this saves a lot of time

Today we’re launching micro1’s PII transformation model, flow-transform 1.0, delivering frontier-level performance across detection, identity synthesis, and transformation of personally identifiable information. On PrivacyBench, our model reaches 96.0% F1, outperforming every detection baseline we tested, including Tonic Textual, Claude Opus 4.8, Sonnet 4.6, Microsoft Presidio, Haiku 4.5 and GLiNER2. Some of the most valuable training data for frontier AI models lives inside fully functioning companies. It captures years of real work across decisions, communications, tools, handoffs, exceptions and the relationships connecting them. The problem is that this data is also full of PII. Traditional redaction makes the data safe, but it also destroys the very workflows and relationships frontier models need to learn from. flow-transform 1.0 solves this by turning enterprise operational data into high-fidelity training data for frontier models by replacing real-world identities without flattening the reality the data captures.

Slash runs on Whop

ONE PERSON CAN NOW RUN A FULL CONTENT TEAM WITH GROK One bot per job. You just make the calls. → Research competitors while you sleep → Pick the one idea worth testing → Build the assets with consistent avatars → Auto-schedule everything in Postiz No camera. No editors. No team. 🔖 Save this. nitter.net/perezhatesai/status/21…

Dreamina AI updated its web experience and canvas, and it’s live now. @Dreamina_ai #DreaminaCaughtTheVibe The update brings high quality AI video workflows and skills for different creative needs, including film, brand ads, viral social media content, and more. It makes it easier to explore ideas and move from concept to content in one place. Get 90% OFF your first month of Dreamina Basic until October 9th. Try it out on your next creative project.

A US financial services company scoped three infrastructure projects for five engineers across two quarters. Our two-person Velocity Pod delivered all three in one quarter: automated funds movement, an underwriting portal, and investor access. I want to explain what made that possible, because “two engineers using AI” leaves out most of the useful information. The platform was already processing live transactions. Existing code, existing integrations, existing customers whose money had to keep moving. An audit found several things preventing the agents from working reliably. Domain rules weren’t documented clearly. Build and test commands were difficult to discover. Different modules followed different conventions. Recurring tasks had no established procedure the agent could follow. So the agents filled in the gaps themselves. That is a pretty expensive place to let software guess. We documented the architecture, domain terminology, coding conventions, and exact verification commands at both repository and module level. We also packaged recurring procedures into reusable skills. Adding an endpoint or writing a migration now had a documented approach. Every task started with a specification. Tests came before implementation. Every change passed automated verification before merge, with a PR review agent checking against the spec and project standards. All three projects used a shared architecture, so each subsequent project started with a foundation already in place. The result: 60% fewer engineers than planned, half the estimated timeline, and about $200 in AI compute per developer per month. The lesson I’d take from this is that your codebase contains less of your company’s knowledge than you think. Your experienced engineers know which conventions matter, where the exceptions live, and why something was built a particular way. An agent needs access to that knowledge too. Making it explicit is part of the engineering work. That’s what our Velocity Framework was built for.

extremely important if you run a fintech company: